Introduction

Mental ill-health causes a substantial burden, both on an individual level and in terms of costs to society and healthcare systems [1, 2]. One step towards reducing the burden is to understand how to break the vulnerability processes leading up to mental ill-health, as well as understanding how to foster resilience processes. The former goal can be achieved by preventing accumulation of risk factors, while the latter goal translates into increasing availability of resources that promote mental health, especially for those at increased risk. If a resource is particularly beneficial for individuals at higher risk, it is a buffer of that risk. Against this backdrop, this study focuses on severe childhood adversity, indicated by out-of-home care (OHC) [3], as predictor of psychiatric disorders across the life course [4], and extra-familial risk and resource factors that might matter for this association, with a particular interest in gender patterns.

Childhood is a period in which violations of the expectable environment, such as dysfunction in the family, can come to shape physiological and behavioural stress responses in ways that are maladaptive [5–7]. A life course framework poses that early adversity can thus accumulate with time and/or trigger chains of risk, which can lead to lasting disadvantage [8]. Previous empirical research on the accumulation of risk following early adversity has identified various pathways to mental ill-health, with poor educational outcomes emerging as one significant factor [4, 9]. Early adversity is related to an increased likelihood of lower educational achievement, which in turn often translates into poorer labour market outcomes [10]. Considering the socioeconomic gradient present in nearly all health outcomes, and the role that the distribution of social and material resources plays in contributing to health inequalities, it is plausible that education serves as a significant mediator between early adversity and later mental ill-health [11]. Relatedly, other extra-familial factors that can interfere with educational achievement, such as early parenthood and delinquency, are relevant to analyses of vulnerability processes, especially from a gender perspective. For example, early parenthood is more common in individuals with experience of childhood adversity [12], and can prevent especially young mothers from establishing themselves in the labour market [13, 14]. Furthermore, motherhood in adolescence increases the risk of poor health outcomes, especially in conjunction with low social support [15–17]. Similarly, delinquency and poor mental health are associated [18, 19], but delinquency is a risk factor that is more frequently observed in boys than girls [20]. It is also more prevalent among those with experience of childhood adversity [21]. Both early parenthood and delinquency therefore constitute promising candidates for examining gendered risk accumulation following adversity.

While early adversity is a strong risk factor, it is not deterministic [8, 22]. Resilience research therefore explores heterogeneity in outcomes after severe stress exposure, and investigates amongst other things potential protective factors that can even out said stress exposures [23]. Resilience refers to multi-domain processes that include individual-level resources, but also relationships and the wider ecological systems theses are embedded in [23, 24]. Since OHC implies disconnection from the family, and thereby restricted access to resources in the home, it is particularly relevant to explore extra-familial resources as buffers of adversity. Previous research indicates that social support by non-family members is such a buffer [25], and possibly more so among women than among men [25, 26]. Leisure time activities have also proven beneficial for young people from disadvantaged backgrounds [27]. Both close friendships and meaningful leisure time activities can provide youth with a sense of connection and belonging that are said to promote resilience [28–30]. Lastly, better educational outcomes—as the flip-side of the risk imposed by poor educational outcomes—represents another potential buffer. On the one hand, education is an overarching resource and a strong predictor of future socioeconomic status [11]. On the other hand, education also provides more specific resources, such as health literacy [31].

There is a considerable amount of research on the pathways along which risks accumulate after exposure to childhood adversity, as well as on factors that buffer the impact of such adversity. In studies of mental ill-health, the former has primarily been investigated by means of mediation analysis, whereas the latter typically is addressed through moderation analyses. Despite well-documented gender patterns of both mental ill-health and the distribution of risk and resource factors [24, 32], there is a lack of research on how gender might influence the interplay between such factors and mental health outcomes following childhood adversity [24, 33]. This is a research gap this study sets out to fill. Furthermore, even though empirical evidence concerning many potential mediators and moderators of adversity exists, this evidence is often inconsistent across studies [33, 34]. Additionally, existing studies are highly heterogeneous. Consequently, there is a general interest in strengthening–or refuting–current evidence on mediators and moderators of childhood adversity in different contexts [34], identifying the pathways along which vulnerability unfolds, as well as the resource factors which can foster resilience. Lastly, there is a need for prospective studies based on community-based samples to complement existing research, which is often limited to retrospective data and non-representative samples. In the current study, we follow a large cohort of individuals from birth up until retirement, leveraging both register and survey data.

1. The aim of the study is to better understand the accumulation of risk (approached as mediation) and the overcoming of risk (approached as moderation) incurred by childhood adversity, with particular attention to gender patterns. We address the following research questions:
2. Do extra-familial risk factors in adolescence (delinquency, early parenthood, grades) mediate the association between OHC experience and psychiatric disorders in adulthood?
3. Do extra-familial resource factors in adolescence (number of best friends, leisure time activities, grades) moderate the association between OHC experience and psychiatric disorders in adulthood?
4. Are there gender patterns in regard to 1) or 2)?

Methods

Results

Table 1 shows that a significantly greater proportion of men and women with OHC exposure were diagnosed with a psychiatric disorder compared to those without OHC exposure. In the total sample and in boys, lower school grades, delinquency, and early parenthood were associated with greater risk of ADS and SM. In girls, school grades were only associated with SM, but not with ADS. The risk of a SM-related diagnosis declined by increasing number of best friends in the total sample and in boys.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Frequencies and percentages of OHC and proposed mediators/moderators of the OHC-outcome association, by gender and outcome status.

https://doi.org/10.1371/journal.pone.0318231.t001

Logistic regression estimates of the associations between the outcomes and mediators and moderators (see S2 Fig and S5a and S5b Tables in S1 File) revealed similar patterns as in Table 1. Additionally, having four or more best friends, and structured leisure time activity at least once per week, were associated with reduced likelihood of ADS in the total sample, but not in the gender-stratified analyses.

Potential accumulation of risk—mediation analyses

The overall and gender-stratified mediation results using the KHB method are presented in Table 2. In the full sample, the KHB analyses showed significant indirect associations between OHC and both outcomes through school grades, delinquency, and early parenthood. When stratifying by gender, the association between OHC and ADS in girls did not operate via school grades, while it did so in boys. Conversely, in boys, early parenthood was only a mediator of the association between OHC and SM, but not between OHC and ADS. For ADS, the included mediators explained 36% of the indirect association with OHC in boys, but only 15.8% in girls. For SM, the absolute difference of mediation percentages achieved by the full model was smaller, but still substantial (37% in boys and 25.6% in girls).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Overall and gender-stratified mediation results.

https://doi.org/10.1371/journal.pone.0318231.t002

Potential buffering factors—moderation analyses

In the two-way interaction analyses (S6a-S8b Tables in S1 File), only number of best friends was a statistically significant moderator, although not in the same direction for both outcomes (Table 3, and S7a and S7b Tables in S1 File). More specifically, while having more friends was particularly beneficial for boys with OHC experience in terms of preventing ADS, it was an exacerbating factor for SM in the total sample (S7a and S7b Tables in S1 File). The results further show that there was a tendency for higher grades to even out the risk introduced by OHC experience for both outcomes (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Predicted probabilities of anxiety, depression, and self-harm, and substance misuse.

Shown across levels of each moderator, by gender and OHC experience.

https://doi.org/10.1371/journal.pone.0318231.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. P-values from likelihood ratio tests.

https://doi.org/10.1371/journal.pone.0318231.t003

Gender did not interact with OHC in relation to either outcome (ADS: interaction term p = 0.74; SM: interaction term p = 0.76). None of the three-way interaction terms (OHC x gender x moderator) were significant (Table 3).

Discussion

The present study analysed the role of extra-familial risk and resource factors in relation to the association between childhood adversity, as indicated by OHC, and psychiatric disorders in adulthood, with a special interest in gender patterns. Such analyses provide opportunities for a deeper understanding of how vulnerability unfolds after OHC, and for identifying resource factors that may foster resilience.

The results show that OHC is prospectively associated with psychiatric disorders in adulthood. Risk appears to accumulate along pathways from OHC to adult psychiatric disorders via low school grades, delinquency, and early parenthood for both genders. The exception were low school grades in girls and early parenthood in boys, both in relation to ADS. Education appeared to be a more important mediator in boys than in girls, similar to findings from previous studies in the context of childhood socioeconomic disadvantage and depression [9, 39]. Thus, the transmission of risk along the educational pathway seems to be stronger for men. Correspondingly, educational success also bears greater protective potential for men. Why this might be the case is discussed in more detail in the context of moderation analysis later on. With regards to SM, school grades mediated a similar proportion of the association across genders, suggesting that gender patterns might be outcome-dependent.

To the best of our knowledge, our study is the first to consider delinquency and early parenthood as possible gendered pathways between OHC and psychiatric disorders. We found clear evidence of this being the case, as delinquency mediated a substantial portion of all investigated associations in boys, but not to the same extent in girls. This is somewhat different to a comparable study in which no marked gender difference in mediation via delinquent behaviour was identified in the context of adverse childhood experiences and non-medical opioid use [40]. Although not directly comparable, due to differences in the study context and outcomes, the divergence in our and Quinn et al.’s findings highlight the importance of replication and context-sensitivity in research on risk accumulation after childhood adversity. Our own findings fit with the general observation that men tend to exhibit externalising problems more frequently than women [e.g., 41], which is in line with gendered expectations around coping behaviours [42]. Conversely, early parenthood was a more important mediator in girls. Again, while we are not aware of other studies investigating exactly the same links, those exploring the most similar exposure-outcome combinations to our study, also identified gender disparities in the influence of early parenthood. For example, one study found a significant indirect effect of childhood socioeconomic disadvantage via early parenthood on heart problems in women, but not in men [14]. Taken together, the evidence suggests that early parenthood is a greater source of risk accumulation in women with experience of childhood adversity compared to men. We argue this may be due to gendered norms around care-taking as well as greater stigma and isolation related to young motherhood when compared to young fatherhood. We argue that the emergence of delinquency and early parenthood as gendered risk accumulation pathways calls for a gender-sensitive engagement with youth at risk of experiencing continued vulnerability. Examples of this could be discouragement of criminal behaviour as a demonstration of masculinity and promoting of shared responsibility in relation to early parenthood.

Overall, a greater portion of the total associations in boys could be explained through the included mediators, implying that for girls, there might be other, more important risk factors after OHC that we could not capture with our data. Drawing on previous empirical research as well as theory on gendered vulnerability processes, such factors may encompass psychosocial aspects, such as coping strategies or feelings of loneliness [40, 43, 44]. In the present study, the only available possible indicator of coping was delinquency, which, as previously noted, is more common, and socially accepted, in boys/men than in girls/women. Overall, the findings on risk accumulation suggest that assisting young individuals in avoiding unplanned parenthood (especially among young women) and delinquent behaviour (especially among young men) represents a gender-sensitive approach to disrupting some of the mechanisms through which childhood adversity can lead to adult psychiatric disorders.

We were not able to clearly identify resource factors that buffer the impact of OHC on psychiatric disorders. Only one of the two-way-interaction terms was significant, namely that between OHC and friendship, but the direction was inconsistent: The results indicate that increasing number of friends among boys evens out the risk of ADS, bringing those with OHC placement on a similar risk level to those without. In the total sample, the reverse was true: a greater number of friends disproportionately increased the risk of SM for those with OHC experience. This points to the importance of context–not only the number, but also the quality of friendships matters for increases or decreases in the risk of negative outcomes [45]. In contrast to existing evidence on the importance of social support for girls and women [e.g., 25, 26], no moderation was observed in girls. One possible explanation is a shift in the positive influence of social connection throughout the life course. During adolescence, gender differences in friendship values intensify, and while girls value psychological provisions such as emotional support more than boys [46], they may also experience more stress around nurturing their relationships [47]. It is possible that for those with a large number of friendships, the pressure to preserve these relationships neutralises their positive potential in girls, while boys tend to face lower expectations around prosocial behaviour in the upkeep of social relationships [48].

Despite the lack of statistical significance, our results indicated that higher school grades may even out the risk of psychiatric disorders. High school grades can be considered an early proxy of positive educational achievement, with good predictive power for career trajectories later on in life [49]. While other studies have repeatedly shown the general value of education as a resource and its role as mediator, studies do not necessarily find it to be a buffer in the context of childhood adversity [50–53]. Our findings also support the view that education is more important in its function as mediating pathway, but still suggest it can also act as buffer, i.e., in this context as a resource that is particularly powerful following childhood adversity. Since it is well established that adverse childhood experiences, even before a clinical problem is manifest, may translate into behavioural problems interfering with school performance [54], it seems reasonable to assume that education is primarily a mediator. However, in cases where a child is enabled to “beat the odds” and still do well in school, despite heightened difficulties in the home environment, they may draw even more advantage from educational success than a child without such experiences. While the present study failed to identify any significant gender differences in moderation, the results suggest that success in school was especially helpful in preventing negative outcomes in boys, which could be due to gendered expectations around schooling. Girls typically experience more stress in relation to school performance [55]. This could constitute a counterweight to the benefit of greater access to resources provided by higher education, at least in relation to ADS. Alternatively, even though women in this cohort entered the workforce in larger numbers than previous generations, school grades may still have been more directly linked to future SES in boys. This is may be due to, for example, more limited career opportunities for women, with persisting horizontal and vertical labour market segregation by gender [56].

Lastly, in contrast to previous research [27], leisure time activities in adolescence did not act as a buffer in the association between OHC and adult psychiatric disorders in our data, apart from possibly in boys in relation to ADS. This lack of a clear association may be due to low statistical power, and that leisure time activities were limited to organised club memberships. Future studies should consider both structured and unstructured forms of leisure time activity.

Conclusions

This study shows how extra-familial risk factors accumulate along pathways linking OHC to adult psychiatric disorders. This occurs in an outcome- and gender-dependent fashion, with early parenthood being a particularly important risk factor in girls with OHC experience, and delinquency a particularly important risk factor in boys. We found limited evidence for friendships and higher school grades acting as extra-familial resource factors that even out the excess risk of psychiatric disorders among those exposed to OHC. In conclusion, there is much to suggest that vulnerability and resilience can be understood as context-dependent and gendered processes, the details of which warrant further investigation.

Supporting information